1 |
Am Thu, 08 Feb 2018 14:50:31 -0500 schrieb Rich Freeman: |
2 |
|
3 |
> On Thu, Feb 8, 2018 at 2:17 PM, Dale <rdalek1967@×××××.com> wrote: |
4 |
>> As someone else pointed out, if you start using swap, that generally |
5 |
>> defeats the purpose of tmpfs. |
6 |
>> |
7 |
>> |
8 |
> I'll just add one thing to this, which I've probably already said ages |
9 |
> ago: |
10 |
> |
11 |
> In an ideal world swap would STILL be better than building on disk, |
12 |
> because it gives the kernel fewer constraints around what gets written |
13 |
> to disk. |
14 |
> |
15 |
> Anything written to disk MUST end up on the disk within the dirty |
16 |
> writeback time limit. Anything written to tmpfs doesn't ever have to |
17 |
> end up on disk, and if it is swapped the kernel need not do it in any |
18 |
> particular timeframe. Also, the swapfile doesn't need the same kinds of |
19 |
> integrity features as a filesystem, which probably lowers the cost of |
20 |
> writes somewhat (if nothing else after a reboot there is no need to run |
21 |
> tmpreaper on it). |
22 |
> |
23 |
> So, swapping SHOULD still be better than building on disk, because any |
24 |
> object file that doesn't end up being swapped is a saved disk IO, and |
25 |
> the stuff that does get swapped will hopefully get written at a more |
26 |
> opportune time vs forcing the kernel to stop what is doing after 30s (by |
27 |
> default) to make sure that something gets written no matter what (if it |
28 |
> wasn't deleted before then). |
29 |
|
30 |
I can only second this. |
31 |
|
32 |
> That's all in an ideal world. In practice I've never found the kernel |
33 |
> swapping algorithms to be the best in the world, and I've seen a lot of |
34 |
> situations where it hurts. I run without a swapfile for this reason. |
35 |
> It pains me to do it because I can think of a bunch of reasons why this |
36 |
> shouldn't help, and yet for whatever reason it does. |
37 |
|
38 |
I really prefer having inactive things being swapped out than discarding |
39 |
cache from memory. But since kernel 4.9 this no longer works so well. I'm |
40 |
still seeking the reason. But for that reason, building in tmpfs is no |
41 |
longer such an appealing option as before. |
42 |
|
43 |
Otherwise, I was quite happy with swap behavior, exactly for the reasons |
44 |
you initially outlined. |
45 |
|
46 |
|
47 |
-- |
48 |
Regards, |
49 |
Kai |
50 |
|
51 |
Replies to list-only preferred. |